Advancing Spatiotemporal Prediction using Artificial Intelligence: Extending the Framework of Geographically and Temporally Weighted Neural Network (GTWNN) for Differing Geographical and Temporal Contexts

TL;DR

This work extends the Geographically and Temporally Weighted Neural Network (GTWNN) framework to address spatiotemporal crime prediction across London and Detroit by introducing multiple architectural extensions, including intermediate layers, spatiotemporal output expansion, and a history-dependent module. Bayesian optimization NAS is used to search architectures, and an isotropy-based data augmentation check informs preprocessing choices. Empirical results reveal that the history-dependent module is particularly beneficial when temporal correlations are strong, whereas spatial continuity-focused variants excel where spatial correlation dominates; certain combinations can be redundant. The study provides practical guidance on selecting GTWNN variants based on dataset-specific PACF structure, contributing to more context-aware, accurate spatiotemporal predictions in crime analytics and, more broadly, spatiotemporal modelling.

Abstract

This paper aims at improving predictive crime models by extending the mathematical framework of Artificial Neural Networks (ANNs) tailored to general spatiotemporal problems and appropriately applying them. Recent advancements in the geospatial-temporal modelling field have focused on the inclusion of geographical weighting in their deep learning models to account for nonspatial stationarity, which is often apparent in spatial data. We formulate a novel semi-analytical approach to solving Geographically and Temporally Weighted Regression (GTWR), and applying it to London crime data. The results produce high-accuracy predictive evaluation scores that affirm the validity of the assumptions and approximations in the approach. This paper presents mathematical advances to the Geographically and Temporally Weighted Neural Network (GTWNN) framework, which offers a novel contribution to the field. Insights from past literature are harmoniously employed with the assumptions and approximations to generate three mathematical extensions to GTWNN's framework. Combinations of these extensions produce five novel ANNs, applied to the London and Detroit datasets. The results suggest that one of the extensions is redundant and is generally surpassed by another extension, which we term the history-dependent module. The remaining extensions form three novel ANN designs that pose potential GTWNN improvements. We evaluated the efficacy of various models in both the London and Detroit crime datasets, highlighting the importance of accounting for specific geographic and temporal characteristics when selecting modelling strategies to improve model suitability. In general, the proposed methods provide the foundations for a more context-aware, accurate, and robust ANN approach in spatio-temporal modelling.

Figures (21)

• Figure 1: Pictorial representation of the vanilla NN model applied to crime prediction.
• Figure 2: Pictorial representation of the GWANN model applied to crime prediction.
• Figure 3: Pictorial representation of the GTWNN model applied to crime prediction. The circle-dot symbol represents element-wise multiplication.
• Figure 4: Pictorial representation of the hybrid model, GTWNN_Ls, applied to crime prediction.
• Figure 5: Pictorial representation of the GTWNN_Lst model applied to crime prediction.